Married men make a lot more money than single men.  In the NLSY, married men make 44% extra, even after controlling for education, experience, IQ, race, and number of children.  How is this possible?

There are three competing economic explanations.  Each of the three may be partly true.

Explanation #1: Ability bias.  The causal effect of marriage on male income is smaller than it seems.  Even after adjusting for all the previously listed control variables, men with higher income are simply more likely to be married.  Maybe income makes it easier to attract a spouse; maybe Puritan attitudes lead to both income and marriage.  In a pure ability bias story, marriage has zero causal effect on earnings.

Explanation #2: Human capital.  Marriage causally increases male income by making men more productive workers.  Maybe marriage makes men work more hours; maybe it makes them work harder per hour; maybe it makes them control their tempers better; maybe all of these and more.  In a pure human capital story, marriage actually causes men to become 44% more productive.

Explanation #3: Signaling.  Marriage causally increases male income by changing employers’ beliefs about worker productivity.  As long as married men happen to be more productive, and employers can’t costlessly see their productivity, employers will rationally (and profitably!) pay married men more.  In a pure signaling story, marriage makes employers expect you to be 44% more productive, but has zero causal effect on productivity.

We can summarize these competing explanations with a table:


Causal Effect on Productivity

Causal Effect on Employers’ Beliefs About Productivity

Causal Effect on Income













Economists who study the male marriage premium usually conclude that much of it is causal.  This paper, for example uses shotgun weddings to isolate the causal effect of marriage on income, and finds:

Using the statistical experiment of premarital conception as a potentially exogenous cause of marriage, about 90% of the marriage premium remains after controlling for selection.

If shotgun weddings are genuinely exogenous, we can use them to measure the causal effect of marriage.  But as the preceding table indicates, there are two competing causal stories.  And they’re very hard to empirically distinguish.  A shotgun marriage could causally increase your earnings by improving your attitude on the job.  But a shotgun marriage could just as easily causally increase your earnings by showing employers that you belong to a category of workers – married men – that typically have a good attitude.  “Selection,” properly interpreted, refers to ability bias alone, not (ability bias and signaling).

If that’s unclear, consider the case of tattoos.  A facial tattoo could causally reduce your income by giving you a bad attitude; but it could just as easily (and more plausibly) causally reduce your income by showing employers that you belong to a category of workers – guys with facial tattoos – that typically have a bad attitude.  A study of “shotgun tattoos” could tell you if the tattoo penalty were causal, but couldn’t empirically distinguish the human capital from the signaling mechanisms.

So what is the male marriage premium?  I’m still deciding, but here’s my tentative opinion.

1. The shotgun wedding paper notwithstanding, I think that about half of the marriage premium stems from ability bias.  Men who marry are just more conscientious, ambitious, and cooperative, and the NLSY lacks good measures of these traits.  This remains true even when men have a shotgun wedding; the stand-up guys go through with the wedding, while the slackers skulk away.

2. At least in the modern American economy, the signaling channel explains no more than 10% (not 10 percentage-points) of the male marriage premium.  My reasoning: When employers make hiring decisions, they heavily scrutinize educational credentials, but barely notice marital status.  I can easily believe that the signaling channel was far more important in the past; when almost every man marries, the failure to marry raises a red flag.  But nowadays?

My main doubt is that I know little about hiring in more traditional occupations and regions of the country.  Do employers in Kansas still raise their eyebrows when they see that a 35-year-old male applicant is single?  What about CBN?

3. If the male marriage premium is 50% ability bias, and less than 10% signaling, then human capital explains the rest: 40-50%.  Much of this effect probably reflects longer work hours and lower unemployment.  But it’s quite plausible that marriage causally increases hourly productivity by 10%.

Is my breakdown correct?  If not, what’s the correct breakdown between ability bias, human capital, and signaling?  Please show your work.

P.S. Coming soon: “What Is the Female Marriage Premium Penalty?”